No label defined (Q4933739)

The book is devoted to an introduction to the analysis of diffusion paths on a Riemannian manifold, emphasizing the geometric aspect rather than the stochastic aspect of the topics. The author tries to minimize the weight of probabilistic tools and wrote an excellent book, very accesible to readers who are not accustomed with probability (for instance, Stratonovich calculus is used instead of Ito's stochastic calculus). Chapter 1 is devoted to Lévy's construction of the Wiener measure and to description of some properties. In Chapters 2 and 3, diffusions more general as Brownian motion are viewed as integral curves [see also \textit{D. Stroock} and \textit{S. Taniguchi}, Prog. Probab. 34, 333-369 (1994; Zbl 0814.60074)]. In Chapter 2 everything is sufficiently bounded so that no problems about explosion arise and in Chapter 3 it is shown that much continues to hold even after the boundedness assumptions are removed. Diffusions on manifolds (Chapter 4) can be viewed as special cases of the diffusions constructed in Chapters 2 and 3. When \(M\) appears as an embedded submanifold of \(\mathbb{R}^{N}\), Brownian motion on \(M\) can be realised by projecting Wiener paths from \(\mathbb{R}^{N}\) onto \(M\). Extrinsic ideas are contained in Chapters 5 and 6. Curvature considerations are introduced in Chapter 5: the Ricci curvature explains some behavior of Brownian paths (in connection with Yau's non-explosion criterion). In Chapter 6, the Bochner identity is proved in an integrated form [see also \textit{J.-M. Bismut}, `Large deviations and the Malliavin calculus' (1984; Zbl 0537.35003)]. The rest of the book contains intrinsic ideas. In Chapter 7, it is explained how we can transform the ideas of Chapters 2, 3 and 4 to the setting of an abstract differentiable manifold (for instance, a construction of Brownian motion via localization). Chapter 8 contains basic facts about the bundle \({\mathcal O}(M)\) of orthonormal frames and gives a construction of Brownian motion (and properties) as a projection to \(M\) of the diffusion on \({\mathcal O}(M)\) associated with Bochner's Laplacian [see also \textit{J. Eells} and \textit{K. D. Elworthy}, Control Theory Top. Funct. Anal., Vol. III, 179-185 (1976; Zbl 0355.60053) and \textit{P. Malliavin} and \textit{D. Stroock}, J. Differ. Geom. 44, No. 3, 550-570 (1996; Zbl 0873.58063)]. The idea of Chapter 9 is to expose how normal coordinates are used to the study of Brownian motion on a manifold. The applications are strictly local (exit time, exit place of Brownian motion from small balls). The final chapter [derived from \textit{O. Enchev} and \textit{D. W. Stroock}, J. Funct. Anal. 134, 392-416 (1995; Zbl 0847.58080)] contains the so called Malliavin calculus: useful analytic information can be obtained doing differential calculus in path space [see \textit{S. Kusuoka} and \textit{D. Stroock}, Stochastic analysis, Proc. Taniguchi Int. Symp., Katata \& Kyoto/Jap. 1982, North-Holland Math. Libr. 32, 271-306 (1984; Zbl 0546.60056); J. Fac. Sci., Univ. Tokyo, Sect. I A 32, 1-76 (1985; Zbl 0568.60059); J. Fac. Sci., Univ. Tokyo, Sect. I A 34, 391-442 (1987; Zbl 0633.60078); Ann. Math., II. Ser. 127, No.1, 165-189 (1988; Zbl 0699.35025) in the Euclidean setting and see \textit{J.-M. Bismut} (loc. cit.) in the differential geometric setting; see also \textit{B. K. Driver}, J. Funct. Anal. 110, No. 2, 272-376 (1992; Zbl 0765.60064)].